Post by QPerfect

🔬⚛️ What if doubling your atom-loss events could actually boost the performance of your quantum error-correcting code? It sounds backwards. In our previous work, we showed that properly accounting for atom loss during decoding can significantly improve the performance of quantum error correction in neutral-atom processors. Now, our latest paper reveals there's even more information hidden in those loss events: the correlations between them. 📄 Previous paper: https://lnkd.in/eEbHVVFh 📄 New paper: https://lnkd.in/e3wpYdhP Atom loss is the dominant error source in neutral-atom processors, arising mainly during the Rydberg-mediated CZ gates that entangle pairs of atoms. Here's the key physical fact: when one atom is lost during the gate pulse, it can re-excite its partner to the Rydberg state — sharply increasing the probability that the partner is lost as well. These paired-loss events are experimentally ubiquitous, yet existing decoders typically treat every loss as independent. For the rotated surface code equipped with teleportation-based loss-detection units, exploiting these correlations yields: ✅ Up to a tenfold reduction in logical error rate ✅ Loss threshold raised from 3.2% to 4% ✅ A fully parallelizable decoder running in the sub-millisecond range — compatible with real-time operation ✅ Robust gains in the experimentally relevant partially correlated regime, with clear improvement already above 50% correlation ⚙️ How does it work? Our decoder builds a "loss graph" — nodes are lost atoms, edges are the physical channels that lose pairs together — and dynamically updates the probability of each loss channel from the observed syndrome. This converts an ambiguous, delayed erasure (you know an atom is gone, but not exactly when) into a precisely located one (you know where the error is). More broadly, it's a case study in why noise-tailored, architecture-specific decoding matters as neutral-atom platforms move toward fault-tolerant operation. This work is a collaboration between QPerfect and the European Center for Quantum Sciences (CESQ) — another example of industry and academia advancing scalable, fault-tolerant quantum computing together. ⚛️ Atom loss isn't just something to survive — it's information you can exploit. Curious to hear what the neutral-atom community thinks 👇 QuEra Computing Inc. Pasqal Atom Computing planqc Oratomic